Discourse Analysis

There is a wide range of theory and method that uses the term discourse analysis (henceforth DA), largely because the varieties of DA derive from different academic disciplines. All of them are concerned with the structures and functions of discourse, or talk and text. However, most varieties of DA have little obvious connection with each other. We focus here on the kind of DA that is closely tied to the analysis of social actions and interaction, and that has developed primarily within the social sciences, and social psychology in particular. This excludes the cognitive psychology of “discourse processing,” as well as developments within linguistics that extend grammatical and pragmatic analysis beyond the boundaries of single sentences. An impression of the broad range of kinds of DA can be gleaned from Teun Van Dijk's (1985) four-volume collection, as well as from other entries in this encyclopedia, particularly those on critical discourse analysis, which is grounded in linguistics; conversation analysis, which is based in sociology; and narrative analysis, which is derived from linguistics and literary criticism.

In their influential work Discourse and Social Psychology, Jonathan Potter and Margaret Wetherell (1987) defined a theoretical and methodological approach to discourse that focuses on every day descriptions and evaluations, or “versions,” of things. Versions have three major features: construction, function, and variability.

1.
Construction has two senses. Discourse is both constructed (built from ready-made linguistic resources) and constructive (offering particular versions of the world, as distinct from alternative versions).
2.
Discourse is always performative, or functional. That is, things said or written in everyday talk and text are invariably produced with regard to some context of interaction or argument, where they perform actions such as evaluating, criticizing, requesting, confessing, claiming, defending, refusing, and so on.
3.
Versions are variable, which is to say that we should not expect people to be consistent. It is this latter property, variability, that generated a profound critique of traditional theory and methods concerning the nature and measurement of attitudes, and also the conduct and interpretation of interviews.

The three features are mutually implicated. Discourse is constructive of whatever it describes, in that an indefinitely extensive range of alternative descriptions is always possible, without their becoming simply or objectively false. This permits the choice of any particular description to be functional, or action-performative, not only in clear cases, such as requests and questions, but also in the case of ostensibly simple, straightforward, factual descriptions. Indeed, DA's major analytic concern has been with the constructive and functional work done by factual descriptions. Furthermore, given the functional work that versions perform, variability across versions, even within the talk of the same speaker, can be expected on the basis that different versions may be doing different things on different occasions. Again, this principle has generated far-reaching critiques of other methods, including traditional uses of interviews and questionnaires, in which variability is systematically avoided or removed in favor of defining a person's consistent attitude or understanding.